Multi-task learning for cross-platform siRNA efficacy prediction: an in-silico study

BMC Bioinformatics

Table 3 Comparison between linear ridge regression and support vector regression for single task siRNA efficacy prediction.

Test	RMSE
	T1	T2	T3	T4	T5	T6	T7
Linear ridge regression	23.5544	23.0751	12.8477	30.2501	27.8395	32.8025	32.9677
SVR with linear kernel	23.6965	22.1477	13.3903	31.9928	26.1998	32.8823	32.2824
SVR with radial basis function kernel	29.6775	24.4753	13.5664	31.1238	37.2164	36.2681	43.4349
	T8	T9	T10	T11	T12	T13	T14
Linear ridge regression	26.5710	13.6068	13.4394	36.9945	33.6679	17.3333	28.7044
SVR with linear kernel	27.0521	15.2284	25.9767	34.9588	32.8858	19.9620	30.7536
SVR with radial basis function kernel	25.6995	43.3165	25.9767	32.9811	26.6623	19.9620	25.8301

"E" denotes "Experiment". Linear ridge regression and support vector regression(with linear kernel and radial basis function kernel) are trained with 50% of the data from each experiment, respectively. p-value calculated by pair t-test on linear ridge regression and SVR with linear kernel is 0.2592. p-value calculated by pair t-test on linear ridge regression and SVR with radial basis function kernel is 0.0913.

ISSN: 1471-2105